Isolation anomaly quantification through heuristical pattern detection

Abstract

In an approach for calculating a probability of a consistency violation of a transaction in a database management system, a processor receives a plurality of transactions within a predetermined time period. A processor identifies a first pattern in a first transaction of the plurality of transactions based on at least an isolation level of the transaction. A processor identifies a second pattern, wherein the second pattern is the cooperation between the first transaction and the second transaction. A processor determines that the first pattern of the first transaction substantially matches a preexisting pattern, wherein the preexisting pattern corresponds to values for a mathematical model for estimating a percentage of transactions in violation of consistency criteria. A processor extracts values from the first transaction based on the determined preexisting pattern. A processor calculates the mathematical model using the values from the first transaction and the values from the preexisting pattern.

What is claimed is: 
     
       1. A method for calculating a probability of a consistency violation of a transaction in a database management system, the method comprising:
 receiving, by one or more processors, a plurality of transactions within a predetermined time period; 
 identifying, by one or more processors, a first pattern in a first transaction of the plurality of transactions based on at least an isolation level of the transaction; 
 identifying, by one or more processors, a second pattern, wherein the second pattern is the cooperation between the first transaction and a second transaction within the plurality of relevant transactions; 
 determining, by one or more processors, that the first pattern of the first transaction matches a preexisting pattern, wherein the preexisting pattern corresponds to values for a mathematical model for estimating a percentage of transactions in violation of consistency criteria; 
 extracting, by one or more processors, values from the first transaction based on the determined preexisting pattern; and 
 calculating, by one or more processors, the mathematical model using the values from the first transaction and the values from the preexisting pattern. 
 
     
     
       2. The method of  claim 1 , further comprising:
 recording, by one or more processors, a result of calculating the mathematical model using the values from the first transaction and the values from the cooperation between the first transactions and the second transaction within the plurality of relevant transactions and the values from the preexisting pattern in a storage location. 
 
     
     
       3. The method of  claim 1 , wherein the mathematical model is a function of at least the isolation level of the first transaction, time gap between consecutive transactions from a connection, and average hotspot size. 
     
     
       4. The method of  claim 1 , further comprising:
 calculating, by one or more processors, an average probability of a consistency violation of a transaction in a database management system for the plurality of transactions based on at least a result of calculating the mathematical model using the values from the first transaction and the values from the preexisting pattern and a set of one or more previously calculated results, wherein the previously calculated results are each associated with a distinct plurality of transaction. 
 
     
     
       5. The method of  claim 1 , further comprising:
 pruning, by one or more processors, one or more non-relevant transactions from a plurality of hot spots based on the first pattern of the first transaction. 
 
     
     
       6. The method of  claim 1 , wherein the one or more non-relevant transactions each have a pattern different from the first pattern of the first transaction. 
     
     
       7. The method of  claim 2 , further comprising:
 receiving, by one or more processors, a second transaction, wherein the second transaction and the first transaction have a similar first pattern; and 
 applying, by one or more processors, the result of calculating the mathematical model to the second transaction based on the first pattern of the second transaction, wherein the first pattern of the second transaction is applied to the plurality of relevant transactions to form a second pattern.